Data governance
Governed data as a capability, not a cleanup project.
I architected and lead the enterprise commercial data-governance framework and single source of truth a global rare-disease pharma organization runs on. Before that, I built its global commercial reporting ecosystem from the ground up.
Why a cleanup fails, the three things that make a number trustworthy, and how I run it - below.
Why a cleanup project fails, and a foundation does not
Reporting and analytics across a global brand portfolio were being rebuilt team by team and source by source. The real cost is not the duplicated effort. It is that leadership cannot tell which number is right, so decisions wait or get made on the loudest dashboard. A cleanup project does not fix that, because a cleanup decays the moment it ships. A governed foundation does, because ownership, standards and lineage are built into the platform rather than maintained by goodwill.
Three things have to be true for people to trust a number
Lineage
Every field traceable to its source, so analytics are auditable. When a number is challenged you can show where it came from instead of arguing about it.
Quality and standards
Standards, ownership and quality rules defined once and enforced with IT. A rule that lives in a slide deck is not a rule.
Stewardship
Dictionaries, documentation and enablement so teams self-serve on governed assets. The real measure of a foundation is how many people build on it without asking me.
How I run a governance program
- Start from the decisions the business is trying to make, not from a catalog of tables. Trying to govern everything at once is how governance programs die.
- Pick owners before you pick standards. A standard without a named owner is a suggestion.
- Build the enablement at the same time as the platform. If the documentation and the training arrive six months after the tables do, teams have already built their own version and you are now negotiating.
- Measure adoption, not coverage. The count of governed tables is a vanity metric. The count of teams who stopped maintaining their own copy is not.
The hard part was people, not the platform
I led across dozens of people in internal marketing, sales, platform management and IT, and I managed external marketing and development agencies, all without formal reporting lines. I delivered the governed foundation in six months, and it has been extended and enhanced since. Reporting and analytics are now consistent and reusable across the global brand portfolio, instead of every team rebuilding its own version.
It consolidated dozens of data sources for hundreds of people now building on the governed data. Those figures are scope, not outcome. The outcome is that a brand team in one region pulls a number and a brand team in another region gets the same one. It runs on Databricks and Salesforce Data Cloud as a medallion lakehouse of reusable data products, with lineage and stewardship, worked in Python and SQL.